The global demand for end-stage hip and knee osteoarthritis surgical treatment is rising, as is the need for optimal postoperative rehabilitation.
Patient stratification is key to providing rehabilitation professionals and policymakers with real-life data supporting early discharge planning and continuous care provision.
IRCCS Ospedale Galeazzi-Sant’Ambrogio and the PREPARE Rehab Group led a retrospective, observational study on a population who underwent total hip or knee replacement and subsequent inpatient rehabilitation due to primary or secondary osteoarthritis in a high-volume orthopedic hospital in Milan, North of Italy, testing what variables could be predictive of inpatient rehabilitation length of hospital stay, among a set of clinical, demographic, social and organizational dataset.
As one of the most important goals of the PREPARE project is to train, test, and validate a predictive model of rehabilitation outcomes based on the specific area of each clinical partner, the goal of this study was to run preliminary testing of the model predictivity, and seek for potential further variables which may enhance it.
The Principal Investigator constructed a patient-centered dataset to link various information available to the hospital, either from paper scanned clinical records or administrative databases. A final dataset including 45.306 variables from 1678 complete patient files was analyzed.
The primary outcome of the study was the length of inpatient rehabilitation stay, while the secondary outcome was the burden of care needed by the patient from the caregiver at discharge. This choice was carefully evaluated after discussing it with leading inpatient rehabilitation clinicians (chief physician and chief physical therapist). Considering that
- The rehabilitation program is widely standardized, regardless of the specific characteristics of each patient;
- Iron diet, pharmacological pain management, and deep venous thrombosis prevention vary automatically depending on certain parameters (hemoglobin levels, patient-reported pain, number of days after surgery, and major/minor surgery);
- Pharmacological management of pre-existing chronic conditions is beyond the control of the inpatient rehabilitation physician;
- Most patients are elderly individuals who need caregivers ready to support them after discharge;
The predictive value of the model would not rely on which rehabilitation intervention works best on what type of patient; rather, it would rely on the time needed by different types of patients to be ready for a safe discharge, given the same rehabilitation program. This is fundamental to enable early discharge planning by estimating a) when the patient will need support outside of hospital; b) the degree of support needed from the caregiver (if any) to help with basic functional autonomy and the activities of daily living (e.g., gait and use of walking aids).
Considering that isolation is itself a relevant cause of preventable mortality, the fragmentation of postoperative rehabilitation and the promotion of integrated rehabilitation pathways between the hospital, intermediate and primary care, the social capital of the patient (place of living, availability of a caregiver, the frailty of the caregiver or not) and the place of discharge of the patient are considered among baseline characteristics and outcomes.
All variables were initially tested individually using a linear regression model for inpatient rehabilitation stay and a logistic regression model for burden of care at discharge. Variables that resulted significant (p<0.05) were subsequently considered in a single, comprehensive linear regression model, or a single, logistic regression model, respectively.
Age, living with a family, occupational status, baseline Barthel Index, and duration of surgery were predictors of inpatient rehabilitation stay and burden of care at discharge. Sex, primary or secondary osteoarthritis, American Society of Anesthesiologists score, body mass index, transfusion, biological risk, type of anesthesia, day of surgery, numeric pain rating scale, and cognitive function at baseline were not.
These results were published on December 2024, 21 on Applied Sciences. Then, the Principal Investigator discussed with other researchers what variables could enhance the predictive power of the model. Including specific patient comorbidities, surgical access technique, and chronic use of pharmacological therapy can improve the model’s predictive power – collecting this information and testing for predictivity is currently being implemented by hospital researchers.
Authors: Federico Pennestrì, Carlotte Kiekens, IRCCS Ospedale Galeazzi-Sant’Ambrogio